I am designing some particles that will eventually be used over white (or off white).
I had Override Background turned on in Krakatoa, and I had the color set to white. My particles looked particularly contrasty, with home highlights getting brighter, and the shadows getting much darker, where some particles looked black.
I started messing with the lighting, and tried cranking up the emission as well, but the shadows still looked black. But then I saved out an EXR (rendered premultiplied over black by turning off the override) and loaded it into Fusion. Here (after applying a 2.2 gamma) it merged nicely over a white background.
So I am wondering if perhaps the lighting density is somehow being applied incorrectly when you have a n overridden background color. I am not sure if it is an additive/subtractive merge thing, or a linear workflow gamma application thing. But something seems amuck when you set the Override BG Color to white. The output looks different when it is set to black.
It is as expected.
Krakatoa is saving EXRs with Pre-multiplied Alpha. The only background that makes sense in that case is Black. Anything else blends part of the background color into the RGB channels. I am pretty sure I wrote a PDF for another customer a few months ago, will see if I can find it…
The Background override is for the cases you want to do the “compositing” directly in Krakatoa. Or if you want the background to have a non-zero Alpha. Or if the background is already set to a color or bitmap in the Max dialog and you want to make it black instead without making changes there.
I get the premultiplied bit, and why overriding the background to white would render out and comp poorly in Fusion…
However, what I am seeing is in the render display window. Simply rendering there with a white background looks terrible.
Perhaps we are talking about the same thing, and this is indeed what is supposed to happen. But I am wondering if we are talking about two different things.
But what I am talking is that these two cases look different:
A) Set BG to black; render to EXR; Gamma 2.2; Comp over white in post
B) Override BG to White; View the render in the render window in Max
Shouldn’t those two cases look the same?
C ) Override BG to white; render to EXR; try to comp. <-- this I expect to look bad.
I am not sure what you are seeing. Is you Max also set to Gamma 2.2? Even then, the results looks good to me, black or white background.
Start posting screenshots and let’s talk…
Hard for me to reproduce the exact thing I had because I have changed the look so much since then… But some further tests reveal that what is shown in the render window is A below, while I was comping with B below.
A)
Loader -> Gamma 2.2 -> Merge over White
B)
Loader -> Merge over White -> Gamma 2.2
B is the more technically correct, I suppose, since the merge is done in linear space… But it makes the white bloom out over things and looses a lot of density to the particles.
Is Max (or are you) applying the inverse gamma of .4545 to the BG color before the comping? Not that it matter with pure white, of course., which is the issue.
I rendered and saved to EXR both the particles in front of black and in front of white. Krakatoa produces HDRI floating point images and does not touch the Gamma. I loaded both images in Fusion, merged the black background image over white and compared the two.
They were identical.
I added Gamma 2.2 after the merge and compared to the white background image with Gamma 2.2 and they were still identical.
I would not trust the Max Rendered Image Window (aka VFB) to show correct data, and I don’t trust Max’ Gamma controls in general.
But the EXR data goes out of Krakatoa with Gamma 1.0 and nothing gets clamped anywhere, so anything you do in Fusion / Nuke etc. should be fair game.
I could record a Camtasia video of what I am seeing in Fusion. I don’t see the problem you are seeing.
In the following screenshot (scaled down 50%), the A is Black BG merged over White BG, then CC Gamma 2.2 compared with B which is White BG out of Krakatoa, then CC Gamma 2.2. There are absolutely identical.
